Advances in Pb-free Solder Microstructure Control and ... · heterogeneities, and rapid...

Advances in Pb-free Solder Microstructure Control andInterconnect Design

Kathlene N. Reeve, John R. Holaday, Stephanie M. Choquette, Iver E. Anderson, and Carol A. Handwerker

(Submitted April 13, 2016; in revised form May 20, 2016; published online June 9, 2016)

New electronics applications demanding enhanced performance and higher operating temper-atures have led to continued research in the field of Pb-free solder designs and interconnectsolutions. In this paper, recent advances in the microstructural design of Pb-free solders andinterconnect systems were discussed by highlighting two topics: increasing b-Sn nucleation in Sn-based solders, and isothermally solidified interconnects using transient liquid phases. Issues in b-Sn nucleation in Sn-based solders were summarized in the context of Swenson’s 2007 review ofthe topic. Recent advancements in the areas of alloy composition manipulation, nucleatingheterogeneities, and rapid solidification were discussed, and a proposal based on a multi-facetedsolidification approach involving the promotion of constitutional undercooling and nucleatingheterogeneities was outlined for future research. The second half of the paper analyzed twodifferent approaches to liquid phase diffusion bonding as a replacement for high-Pb solders, onebased on the application of the pseudo-binary Cu-Ni-Sn ternary system, and the other on aproposed thermodynamic framework for identifying potential ternary alloys for liquid phasediffusion bonding. All of the concepts reviewed relied upon the fundamentals of thermody-namics, kinetics, and solidification, to which Jack Smith substantially contributed during hisscientific career.

Keywords constitutional undercooling, grain nucleation, liquidphase diffusion bonding, Pb-free solder, ternarydiagrams, transient liquid

1. Introduction

Through a scientific career spanning more than 50 years(1953-2007), Dr. Jack Smith was intrigued by investigationsthat linked thermochemistry with the constitution andcrystal structures of metals, compounds, and alloy systems.His analyses of the thermodynamics of binary, ternary, andmulti-component systems were coupled with detailed eval-uations of the elastic constants of a wide range ofmulticomponent materials, ranging from binary lithiumalloys to yttria stabilized zirconia.[1,2] Fortunately, JackSmith was still at the top of his game and up for newchallenges in the early 1990s when one of our co-authors(IEA) sought his collaboration on the mapping and verifi-cation of a previously undiscovered ternary eutectic reactionin the Sn-Ag-Cu system. Using a classic thermodynamicsapproach, Jack’s guidance on review of the literature andpossible interpretation of the limited data set from previous

work were invaluable in their investigation of ternaryeutectic behavior.[3] Now, as we continue addressingchallenges in the Pb-free transition, in controlling themicrostructure in circuit assembly solder joints and indeveloping high temperature solders to replace high-Pbsolders for hierarchical assembly in multi-chip moduleapplications, we have sought to channel the direction andwisdom of Jack Smith, employing the foundations ofthermodynamics and phase equilibria to guide our analyses.

Today’s electronics industry has transitioned to lead-freecomponents in consumer electronic devices due to theknown risks of lead to human health. The Waste Electricaland Electronic Equipment (WEEE) directive, enacted by theEuropean Union (EU) in 2006, requires all consumerproducts being sold within the EU to be Pb-free.[4] Thislegislation is also supported by the Restriction of HazardousSubstances (RoHS) directive, which restricts the use ofseveral other hazardous materials, including Pb.[5,6] Sincethe electronics industry supplies a global marketplace,compliance with EU regulations has driven the transitiontoward a more restricted set of materials in consumerproducts regardless of where they are sold, particularly forsolder alloys used in circuit board assembly and packag-ing.[7] At the same time, new devices and applications aredemanding higher performance solder alloys for assemblythan currently exist, leading to continuing research in newPb-free alloys and interconnect solutions that meet theseenhanced performance challenges. Furthermore, the transi-tion to Pb-free electronics is not yet complete: the RoHSexemptions for high-Pb, high-temperature alloys used inwafer bumping, substrate bumping, and die attach willlikely continue until a practical replacement becomesavailable. There are no drop-in, Pb-free solders which have

Kathlene N. Reeve, John R. Holaday, and Carol A. Handwerker,Purdue University, 701 West Stadium Ave., West Lafayette, IN 47907,USA; Stephanie M. Choquette, and Iver E. Anderson, AmesLaboratory (USDOE), Iowa State University, Ames, IA 50011,USA; Stephanie M. Choquette, and Iver E. Anderson, Iowa StateUniversity, Ames, IA 50011, USA. Contact e-mail:[email protected].

JPEDAV (2016) 37:369–386DOI: 10.1007/s11669-016-0476-91547-7037 �ASM International

Journal of Phase Equilibria and Diffusion Vol. 37 No. 4 2016 369

http://crossmark.crossref.org/dialog/?doi=10.1007/s11669-016-0476-9&domain=pdf

http://crossmark.crossref.org/dialog/?doi=10.1007/s11669-016-0476-9&domain=pdf

solidus temperatures in the 260 to 310 �C range that cansubstitute for high-Pb solders. Based on an examination ofthe binary phase diagrams with one low melting pointelement (Tm< 400 �C), or a low temperature eutectic, andour experience with designing Pb-free alloys to replace theSn-Pb eutectic, it is unlikely such a solder is possible.

From the point of view of processing temperatures, thechallenges in developing both a high performance Pb-freesolder alloy for circuit board assembly and a high temper-ature Pb-free solder or alternative interconnect solution forwafer bumping, substrate bumping, and die attach can beillustrated by examining the use of the Sn-Pb system.Traditionally, solder alloys in both application regimes werecomposed of binary Sn-Pb alloy combinations. These alloyshad many useful properties, including forming a simplebinary eutectic, excellent mechanical properties and ther-momechanical fatigue resistance, good electrical conductiv-ity, excellent wetting characteristics, and a classic eutecticsolidification morphology with two primary solid solutionphases. The large solid solubility of Sn in Pb (18.9 wt.% at183 �C) leads to a range of useful solidus temperatures from183 �C to close to the melting point of Pb (327 �C), and atunable ‘‘mushy’’ (liquid + solid) zone width based on alloycomposition. Soldering is first performed at the chip or dieattach level with high-Pb (Pb-Sn) solders with a high solidustemperature and narrow mushy zone. A standard hightemperature solder composition of Pb-Sn is 95Pb-5Sn(wt.%), with a solidus temperature of 320 �C. Oncesolidified, packaged components are soldered onto circuit

boards without melting the die attach level solder by usingthe Sn-Pb eutectic composition of 62.13Sn-37.87Pb (wt.%),with its eutectic temperature of 183 �C.[8-10]

The microelectronics community has developed a rangeof Pb-free solder alloys for various circuit board assemblyapplications. Tin-based Pb-free alloys, most notably Sn-Ag-Cu (SAC) near-eutectic alloys, discovered by the team thatincluded Jack Smith,[3] have become widely used as analternative for eutectic Sn-Pb due to tin’s affordability,environmental passivity, low toxicity, and compatibilitywith existing microelectronic components and assemblypractice. The SAC eutectic composition is 95.6Sn-3.5Ag-0.9Cu wt.%, with a eutectic temperature of 217 �C.[11] Aprojection of the SAC liquidus surface is shown in Fig. 1(a).After about 20 years of extensive use, the SAC solder alloyshave proven to be a reasonable substitute for Sn-Pb, but stillhave problems in some types of assembly operations, e.g.,ball grid arrays, and more importantly, in terms of theirreliability with respect to thermal fatigue and impactresistance. Interestingly, all of these problems can be tracedto the well-known difficulty in the nucleation of Sn, i.e., incontrolling the as-reflowed joint microstructure. This lack ofmicrostructure control has motivated significant research onthe effects of minor alloy additions to SAC or other binarySn-based alloys on Sn nucleation over a number of years.[12-20] In contrast to circuit board assembly solders in whichmany high-Sn alloys are available commercially, Au-basedeutectics (Au-Sn, Au-Si, Au-Ge) are the dominant Pb-freealloys used for high-temperature soldering applications, but

Fig. 1 Two differing projections of the liquidus surface of the Sn-Ag-Cu ternary system. In (a), the calculated equilibrium liquidus sur-face is shown from Ref 11, and (b) a non-equilibrium projection of the liquidus surface is displayed, showing the extended primaryintermetallic phase formations in the system during eutectic undercooling[21]

370 Journal of Phase Equilibria and Diffusion Vol. 37 No. 4 2016

at significantly higher cost than high-Pb, Pb-Sn alloys.Considered together, this explains the great R&D race todevelop both higher performing Pb-free solder alloys forcircuit board assembly, as well as new approaches orconcepts for widespread high temperature assembly andapplications.

The goal of this article is to provide a review of twoapproaches to designing higher performance solder alloysand interconnect systems: (1) strategies to increase b-Snnucleation in Sn-based solder alloys and (2) liquid phasediffusion bonding (LPDB) to create high temperature, highreliability, Pb-free interconnects. Both of these topics arehighly active research areas in today’s Pb-free soldercommunity and both research areas rest foundationally onconcepts in thermodynamics and phase equilibria, thekinetics of intermetallic formation, and principles of alloysolidification. The first half of this article reviews currentresearch in b-Sn nucleation based on the work publishedsince Swenson’s seminal 2007 review on the difficulties innucleating b-Sn in Sn-based solders.[21] The second half ofthis review focuses on the challenges in designing multi-component LPDB alloys with controlled microstructures asreplacements for high-Pb, high temperature solders.

2. b-Sn Nucleation in Tin-Based Solders

As noted by Swenson in his 2007 review, the solidifi-cation of b-Sn occurs by heterogeneous nucleation and hasbeen difficult to induce or predict, with undercoolings aslarge as 15-40 �C in Sn-based solder alloys, and as large as187 �C in surface passivated Sn droplets.[21,22] These largeundercoolings can permit extended formation of primaryintermetallic compounds (IMCs) far from thermodynamicequilibrium. This results in large, highly anisotropic solderIMCs, such as Cu6Sn5, Ag3Sn, and both Cu6Sn5 and Ag3Snduring solidification of Sn-Cu, Sn-Ag, and SAC alloys,respectively. The eventual nucleation and subsequent rapidgrowth of b-Sn dendrites then follows, resulting in large b-Sn dendrite/grain sizes that frequently exhibit commontwinning misorientations.[23,24]

It is this non-equilibrium, extended proeutectic formationof phases, such as the formation of large, primary Ag3Snblades in SAC alloys, which has dictated the need forimproved impact (‘‘drop shock’’) resistance in Sn-basedsolders. Song et al. performed a telling study on thevibrational fracture properties of SAC solder alloys in 2007,just after the release of Swenson’s review article.[25] Thestudy elucidated the role of Ag3Sn in reducing the ductilityand vibrational dampening of SAC alloys, both of which areimportant properties for high reliability applications in theaerospace and automotive industries. In particular, aciculareutectic Ag3Sn and large, proeutectic Ag3Sn blades wereshown to dramatically reduce solder ductility and adverselyaffect the vibrational performance of the SAC solder alloysby providing preferred pathways for accelerated crackgrowth.[25] By enhancing the nucleation of the b-Sn phaseduring solder alloy solidification, proeutectic formation ofAg3Sn and metastable Ag3Sn far below the eutectic

temperature can be avoided, leading to improved mechan-ical properties.

In addition to the b-Sn nucleation issues, the b-Sn phaseitself is known to exhibit large elastic and thermal expansionanisotropies due to its body-centered tetragonal (BCT)crystal structure. Thus, the behavior of individual joints,such as the thermal fatigue tolerance and resistance toelectromigration and creep, are all linked to the size,orientation, and morphology of the b-Sn dendrites withinSn-based solder joints.[26] It is due to b-Sn’s anisotropicproperties that microstructural non-uniformity joint-to-jointand the resulting variability in mechanical properties of Sn-based solder microstructures can be a large issue.[26,27] Theability to understand, control, and increase heterogeneousnucleation of the b-Sn phase is, therefore, needed to createmore reproducible solder microstructures with improvedthermomechanical properties for the high reliability, highperformance Pb-free solder alloys that are needed in themicroelectronics industry today.

2.1 Swenson’s 2007 Review of b-Sn Nucleation

In 2007, Swenson reviewed the issues linked to thedifficulty in the nucleation of the b-Sn phase in Sn-basedsolders.[21] Swenson offered three potential approaches/tac-tics to alleviate the b-Sn nucleation issue based on conceptsguided by fundamental thermodynamics and kinetics, withthe goal of avoiding the large, proeutectic formation ofsolder IMCs and formation of a few large b-Sn grainsduring solidification.

The first approach outlined by Swenson was to employthe use of non-equilibrium phase diagrams to informdecisions of alloy composition in order to minimizeproeutectic IMC formations in the as-solidified soldermicrostructures. Swenson illustrated this strategy throughthe use of non-equilibrium SAC ternary liquidus surfaceprojections, such as the diagram show in Fig. 1(b). It wassuggested that, if the expected average undercooling isknown, an alloy composition could be chosen to avoid theformation of primary phases. For example, if the averageundercooling at which one would expect b-Sn nucleation isapproximately 15 �C, then the Ag content in the alloyshould be reduced to a point at which the Ag3Sn IMC phaseliquidus surface is at least 15 �C below the equilibrium SACeutectic temperature, thus avoiding primary Ag3Sn forma-tion completely. Effectively, this suggestion attacks one ofthe issues of delayed b-Sn nucleation, primary IMC phasegrowth, but it fails to address the enhancement of b-Snnucleation itself.

The second approach was to introduce b-Sn solidificationcatalysts to enhance the nucleation of the b-Sn phase, thusreducing eutectic undercooling. The addition of Zn wasreviewed, where it is noted that micro-alloying additions ofZn (0.1 wt.% additions to Sn-3.4Ag-0.9Cu wt.% alloys)reduced theb-Sn eutectic undercooling in SAC+Zn alloys andsuppressed large Ag3Sn IMC blade formations within thealloy microstructures, suggesting enhanced b-Sn nucle-ation.[28,29] Inoculation of b-Sn, either from micro-alloyingadditions or from secondary phases, is a promising approach,although an attributable mechanism for the success of such


micro-alloying additions in the inoculation of the Sn-liquidalloy has been notably absent from the literature.

Finally, Swenson suggested the use of rapid solidificationto avoid formation of large primary IMC particles. Rapidsolidification would be expected to reduce, if not completelyeliminate, primary IMC phase formation via the reduction oftime spent in the liquid state during solidification before b-Sn nucleation occurs. Swenson presented results from rapidsolidification experiments that showed IMC refinement, butb-Sn grain dendrite size and misorientation were shown tobe unaffected.[30,31] The employment of rapid solidificationtechniques to modify and reduce the formation of IMCphases from liquid melts is a proven process, and one thatcan be leveraged to improve Sn-based solders. It is vital tonot only establish techniques for enhancing nucleation of b-Sn and controlling b-Sn grain size within as-solidifiedsolders, but to also understand how to maintain and controlthose effects throughout the course of the multiple reflowcycles that are often necessary for microelectronics manu-facturing. In the sections that follow, research progress inexploiting each of Swenson’s approaches in enhancing b-Snnucleation is reviewed, as well as their implications forimproved solder performance. Finally, suggestions forfuture progress in increasing b-Sn nucleation and control-ling b-Sn grain size are outlined.

2.2 Compositional Modifications to Tin-Based Alloys

In accordance with Swenson’s first approach in employ-ing non-equilibrium phase diagrams to inform alloy com-positional choices, as well as the poor impact andvibrational performance of SAC solders attributed to thepresence of primary Ag3Sn, alloy modifications primarilybased on reductions in, or elimination of, Ag from Sn-based

alloys have been researched heavily within the soldercommunity. Suh et al. studied the effects of reduced Agcontent on the fracture resistance of SAC solders, whereSAC105 (Sn-1.0Ag-0.5Cu wt.%) and SAC405 (Sn-4.0Ag-0.5Cu wt.%) were compared via drop impact testing.[32] Theresults of the study found that the SAC105 alloy displayedsuperior drop impact resistance due to the increased bulkcompliance and high plastic energy dissipation of the alloymicrostructure. This result was linked to the reduced Agcontent of the alloy, and thus the reduced formation ofprimary Ag3Sn, as well as the increased phase fraction of b-Sn and the relatively low elastic modulus of the b-Snphase.[32] Suh et al. discussed the effects of Ag content froma thermodynamic standpoint through the use of Fig. 2(a),where reductions in Ag are shown to alter the Sn-IMC tie-lines, thus increasing the phase fraction of b-Sn withdecreased Ag content, in agreement with Swenson’sdescription of such techniques. Figure 2(b) displays theresults of the drop impact testing of the SAC105 andSAC405 alloys. A greater than 10x enhancement of thenumber of drops to joint failure was realized for SAC105 ascompared to SAC405.[32]

In addition to the reduction of Ag content, the reliabilityand mechanical performance of Sn-based solder alloys thatcompletely eliminate Ag from the alloy design have alsobeen examined. Tsukamoto et al. performed a thoroughstudy of Ni-doped and non-Ni-doped Sn-Cu (SC), SAC, andSn-Pb solder ball grid array (BGA) joint systems via the useof high-speed shear impacting testing.[33] The testingshowed that the SC+Ni solder BGAs out-performed allother alloy compositions in shear impact testing. The SACsolder BGAs were found to have the least resistance to shearimpact testing, with cracking initiating at the IMC interfacialbond layer and brittle failure of the solder joints. The

Fig. 2 (a) A schematic of the SAC ternary phase space in the Sn-rich corner. The effect of reduced Ag content is portrayed via the in-creased phase fraction of the b-Sn phase, as seen by the tie-line representation. (b) The drop impact testing results of the study, wherethe superior performance of the SAC105 (Sn-1.0Ag-0.5Cu wt.%) alloy over the SAC405 (Sn-4.0Ag-0.5Cu wt.%) alloy can be seen[32]

(Color figure available online)


addition of Ni to the SC alloy was found to ‘‘smooth’’ themorphology of the interfacial Cu6Sn5 layer between thesolder BGA and the bonded pad.[33] These results agreed wellwith research that has been conducted in the area of Ni-dopedCu-6Sn5 by a number of different research groups.[34-40]

The research has documented the strong cross-interaction ofCu and Ni in solder joints. Typically, Ni is observed tosubstitute on the Cu6Sn5 IMC lattice for Cu atoms, thuscreating (Cu, Ni)6Sn5, which enhances the impact resistanceof the solder IMC interfacial layer by stabilizing thehexagonal form of Cu6Sn5 (g) down to room temperature,negating its transition to the monoclinic structure (g¢) at186 �C. The de-stabilization of this crystallographic trans-formation by Ni additions avoids the volume change thataccompanies the transition of unalloyed Cu6Sn5 IMC phasewith reductions in temperature, a transformation that can leadto cracking in the phase.[37-40] Ni additions were also shownto enhance the planarization of solder IMC (Cu3Sn and (Cu,Ni)6Sn5) interfacial layers, and the cross-interaction of Cuand Ni has been shown to reduce the thickness of the Cu3SnIMC layer during thermal aging.[33,35]

2.3 Inoculation of Tin-Based Alloys

Micro-alloying additions to SAC and SC solder alloys hasbeen an active research area for the past 10-15 years.[12-20]

Anderson and co-workers have examined a wide range ofadditions to SAC alloys; Al (0.05 wt.%), Zn (‡0.21 wt.%),and Mn (‡0.1 wt.%) were each found to reduce b-Snundercoolings to 2-4 �C, and to reduce, or eliminate, theformation of Ag3Sn proeutectic blades. The early studieswere not able to identify the cause of the observed benefitsfrom the micro-alloying additions, but did state that thespecific micro-alloying additions were chosen because oftheir substitutional solubility in Cu and with the goal ofenhanced nucleation of primary Cu6Sn5 from which the b-Sncould then nucleate.[17] However, if such a nucleationmechanism were to exist between Cu6Sn5 and b-Sn, it islikely that researchers would have confirmed it, given thepresence of Cu6Sn5 in majority of solder alloys and solderjoints.

Based on the micro-alloying research performed on Sn-based solders to date, micro-alloying additions of Al appearto be promising. Boesenberg et al. reported the suppressionof the Ag3Sn blade phase at Al contents greater than0.05 wt.%, but less than 0.20 wt.%, and reductions in b-Snundercooling of �50% at Al concentrations 0.25 wt.%within SAC+Al solder alloys. Finally, Boesenberg et al.described the formation of Cu-Al IMC particles within theAl-modified SAC solder samples, which were identified viaXRD measurements as the Cu33Al17 phase. The Cu33Al17particles were found to be buoyant within the liquid-Sn,floating to the top of the examined solder joints duringsolder reflow, and the Cu33Al17 IMC displayed particularlyhigh hardness values (49.1± 2.5 GPa) via nanoindentationmeasurements.[18] Reeve et al. then surveyed the addition ofAl to various SAC and SC alloys, where the volume fractionof the CuxAly phase was shown to vary dependently withthe Al concentration in the alloy, as expected, and the

solidification temperature of the CuxAly phase was identi-fied at 450-550 �C.[20]

Xian et al. have identified the preferred orientationrelationship between Cu6Sn5 and both Cu33Al17 andCu9Al4.

1�210ð ÞCu6Sn5 jj 10�1ð ÞCuxAlyand 0001½ �Cu6Sn5 jj 111½ �CuxAlyThe CuxAly particles were shown by Xian et al. to providenucleating surfaces for the Cu6Sn5 particles within SC+Alalloys, reducing the overall undercooling and size of theCu6Sn5 particles in the systems.[41,42] Similar results havebeen obtained by Reeve et al. over a wide range of coolingrates.[20,43,44] This orientation relationship between theCuxAly and Cu6Sn5 has important implications to soldermicrostructural control due to the high temperature stability(450-550 �C) of the CuxAly phase. If the CuxAly phase canbe uniformly distributed throughout the solder matrix, thenduring reflow cycling of the solder, the CuxAly phase willremain solid in the alloy melt. This high temperaturestability of the CuxAly phase would then provide preferredand persistent heterogeneous nucleation sites for Cu6Sn5during each solidification cycle and an added level of controlover the formation of the Cu6Sn5 phase. Figure 3(a-f)displays various examples of the epitaxial relationshipbetween the CuxAly phases and the Cu6Sn5 phase within arange of SAC+Al and SC+Al alloy compositions.[20,41,43]

Additionally, Sweatman et al. performed a grain refinementsurvey of various micro-alloying additions to pure Sn.[19] Thestudy surveyed several possible grain refining additions topure Sn, and noted that additions of Zn, Mg, and Al allrefined the as-cast grain size of the Sn+X castings. Theaddition of Al displayed the strongest grain refinement bothwithin the as-cast and recrystallized states at additions of 0.3and 0.5 wt.% Al to pure Sn castings. These results indicate apossible link between Al additions, reduced b-Sn undercool-ing observed within the Al-modified alloys, and possibleenhanced nucleation of the b-Sn phase.[19]

Finally, Belyakov and Gourlay have examined hetero-geneous nucleation of b-Sn from XSn4 type IMC, includingPtSn4, PdSn4, and NiSn4.

[45] The introduction of XSn4 typeIMCs, either present as a primary phase or as an interfacialIMC layer, was shown to significantly reduce the under-cooling of the b-Sn phase within the Sn-X alloys. Under-coolings within the alloys were reduced to 3-4 �C, ascompared to pure Sn undercoolings measured at �35 �Cwithin the study. The orientation relationship between the b-Sn phase and the XSn4 IMCs was identified as:

100ð Þb�Snjj 008ð ÞXSn4and 001½ �b�Snjj 100½ �XSn4Despite the potent catalysis of b-Sn by the XSn4 IMC, itwas noted that the b-Sn grain size remained large andunaffected by the presence of the catalytic IMC. Belyakovand Gourlay made note of this and remarked on the need fora combination of effective heterogeneous nucleation sitesfor b-Sn, such as XSn4, and constitutional undercoolingwithin the liquid-Sn to promote grain refinement of the b-Snphase.[45]


2.4 Rapid Solidification of Tin-Based Alloys

Of the suggestions made by Swenson in his review, lesswork has been completed in the area of rapid solidificationof solder alloys. Reeve et al. recently completed a study ofrapidly solidified SC+Al alloys produced via drip atomiza-tion and melt spinning.[43,44] The work considered threedifferent SC+Al alloy compositions, with Al contentsranging from 0.1 to 0.4 wt.%, and cooling rate treatmentsspanning eight orders of magnitude. The work showed therefinement of the CuxAly and Cu6Sn5 IMC phases down tosub-micron ranges through the application of rapid solidi-fication processing (RSP). Ideal initial solidification coolingrates within the range of 103-104 �C/s (water quenching todrip atomization) were identified for realizing simultaneousIMC particle size refinement and a maintained CuxAly/Cu6Sn5 epitaxial relationship during solidification. Fig-ure 4a-b displays the refinement of both IMC phases and themaintained nucleant relationship between the CuxAly andCu6Sn5 phases for Sn-1.41Cu-0.1Al wt.% drip atomizedalloy microstructures from the study.[43] Unfortunately therewas no observed effect on the nucleation of b-Sn due to theemployed rapid solidification methods.

A coarsening study was also performed by Reeve et al.,after initial rapid solidification via drip atomization and meltspinning. After reflow cycling between 20 and 250 �C for 1-5 cycles, both alloy samples displayed little-to-no coarsen-ing of the CuxAly phase, as compared to the initial as-solidified microstructures. However, the Cu6Sn5 particledisplayed coarsening behavior dependent on the tempera-ture stability of the Cu6Sn5 in the alloy composition.Significant Cu6Sn5 particle coarsening occurred in the alloycomposition where the Cu6Sn5 melted completely uponreflow to 250 �C, but little coarsening of the Cu6Sn5 phase

was observed for the alloy where only partial melting of theCu6Sn5 phase occurred during reflow cycling to 250 �C.The epitaxial relationship between the CuxAly and Cu6Sn5IMC was qualitatively maintained throughout the course ofreflow cycling for both samples. These results show promisefor the control of IMC formation within these Al-modifiedalloys, not only after initial solidification, but also through-out subsequent reflow cycling necessary for solder jointmanufacturing.[44]

2.5 Suggestions for Future Improvements in b-SnNucleation

As summarized above, a great deal of work on enhancingthe nucleation of b-Sn within Sn-based solders has beenconducted since Swenson’s 2007 review. Effective ap-proaches in solder compositional variations to control IMCformation, identification of effective micro-alloying addi-tions and heterogeneous inoculants for b-Sn, and techniquesfor rapid solidification of solder alloys have all contributedto the improvement of Sn-based solder joint microstructuralcontrol. However, a singular Sn-based alloy compositionhas yet to be identified that can holistically control: (1)solder IMC formation, (2) b-Sn undercooling, and (3)refinement of b-Sn grain size with multiple non-twinnedorientations. As mentioned by Belyakov and Gourlay intheir investigation of b-Sn inoculant IMC phases,[45] amulti-faceted approach must be taken to solder alloy designto achieve both enhanced b-Sn nucleation and b-Sn grainsize control, including:

1. alloying to promote constitutional undercooling in theSn-liquid, and

2. introducing high temperature inoculant phase/s to pro-

Fig. 3 Various depictions of the epitaxial relationship between Cu6Sn5 and CuxAly IMC within SAC+Al and SC+Al alloys. Micro-graphs (a, b) were collected by the author from work in Ref 20, and (c) is from deep etching experimentation reported in Ref 43. Micro-graphs (d, f) were modified and complied from the work completed in Ref 41 (Color figure available online)


vide multiple, dispersed heterogeneous nucleationsites with specific activity for b-Sn during solidifica-tion.

The effects of constitutional undercooling on solid/liquidtransformations in metallic systems have been described inclassic works of Tiller, Jackson, Chalmers, and Kurz.[46-51]

StJohn et al. recently reviewed the current understanding ofthe effects of constitutional undercooling on nucleation andgrain refinement in alloys.[52] The process of constitutionalundercooling occurs during solidification as solute isrejected from the solidifying phase, resulting in a concen-tration gradient of solute ahead of the solidification frontand a modification to the local temperature of the adjacentliquid. Thus, every point ahead of the solidification front hasa unique solute concentration and a corresponding liquidustemperature that is below that of the equilibrium liquidustemperature of the system, promoting local, compositionallydependent undercooling in the liquid ahead of the solidify-ing phase. Constitutional undercooling thus works toprovide enhanced activation of nucleation sites directlyahead of the solidification front. The ‘‘growth restrictionfactor’’ (GRF) has been established as the most appropriateparameter for quantifying the effect of solute additions onrestricting the growth of the solid interface in an under-cooled liquid:

GRF ¼ mLC0ðk � 1Þ ðEq 1Þ

where mL is the slope of the liquidus, C0 is the alloycomposition, and k is the partition coefficient. A soluteaddition that results in a high GRF promotes a higher rate ofdevelopment of constitutional undercooling in the liquid,slowing the advance of the solid interface into the under-cooled liquid. This allows for more opportunity fornucleation events to activate ahead of the solidificationfront, and thus promotes grain refinement in the alloy.[52]

Seminal research by Easton and StJohn et al. in varioussystems, notably Al alloys, has shown that a high soluteGRF correlates not only to enhanced constitutional under-cooling, but also to reduction in the extent of the ‘‘nucle-ation free zone’’ (NFZ) that is typically established ahead ofa growing solid-liquid interface. In particular, the combina-tion of constitutional undercooling effects and the introduc-tion of heterogeneous nucleant particles has been shown toenhance grain refinement in alloys. For example, thecombination of inoculant nucleation surfaces (TiB2) andexcess solute in the melt (Ti) provides constitutionalundercooling of the Al liquid, as well as potent Alheterogeneous nucleation sites. This two-step solidificationapproach results in an overall grain refinement effect inthese Al alloy systems.[52-56] Such an approach has yet to beapplied to Pb-free, Sn-based solder alloys, but given theneed for enhanced nucleation of the b-Sn phase and theneed to decrease the b-Sn grain size in solder joints,employing multi-part solidification tactics in solder alloysoffers much promise. Values of GRF for varying soluteelement concentrations to pure Sn can be used as a valuableguide in the selection of possible alloy systems to promoteconstitutional undercooling during b-Sn solidification. Itshould be noted, that although the GRF is appropriate to usein terms of down-selecting potential alloying elements thatpromote constitutional undercooling, it does not ultimatelyguarantee a grain refinement effect. As noted by Eastonet al., grain refinement within an alloy will depend not onlyon the degree of constitutional undercooling correspondingto the chosen solute concentration, but also on the thermalprofile in the liquid ahead of the solidification front, thedistribution and potency of inoculant particles in the liquid,the rate of diffusion of the solute species in the liquid (soluteaccumulation between growing grains can significantlyreduce constitutional undercooling), the evolution of latentheat upon solidification, and the flux of heat away from thesolid-liquid interface.[52-54]

Fig. 4 (a) 2-dimensional polished cross-section and (b) 3-dimensonal deep etched sample, depicting the epitaxial relationship betweenCu6Sn5 and CuxAly IMC within a Sn-1.41Cu-0.1Al wt.% drip atomized alloy (estimated cooling rate: 104 �C/s). (Previously unpublishedmicrographs provided by the author) (Color figure available online)


When considering GRFs to select potential soluteadditions to pure Sn, it is noted that Sn-based alloys havegenerally low solute GRF values for various elementaladditions in the Sn liquid at typical micro-alloying levels.Table I displays several solute options and their correspond-ing GRF values at their maximum solubility levels in the Snliquid, as calculated from thermodynamic data gatheredfrom ThermoCalc version 3, TCSLD3 solder alloy data-base.[57] The table portrays the fact that, especially whencompared to the micro-alloying addition of Ti solute to Al(GRF of 36.8 at 0.15 wt.%[53]), the GRF values of thevarious solute additions to the Sn liquid are considerablylower, specifically at low concentrations. The most promis-ing options from the current list include the possibility ofreaching higher values of GRF with increased alloyingconcentrations, for example, a GRF of 86 at 53 wt.% Bi anda value of 29 at 8 wt.% Zn. It is also interesting to note thatthe micro-alloying elements that are generally accepted ashelping to reduce the undercooling of b-Sn, such as Mn, Al,and Co, all have very low GRFs in the alloy liquid,signaling that any observed benefits in reduced undercool-ing within these systems must stem from a mechanism otherthan constitutional undercooling.

Despite some of these limitations, using GRF concepts toincrease the grain refinement of b-Sn may still be possibleby looking outside the typical range of micro-alloyingadditions. For example, Bi concentrations of 22 wt.% wouldachieve a GRF value of 36 in the Sn-Bi liquid, reachinglevels comparable to micro-alloying additions of Ti to Al. Inaddition to decreasing grain size, other microstructuralfactors influence the balance between the need for improvedfatigue life and improved impact resistance in Sn-basedsolder joints. As summarized above, the generally poorimpact resistance of SAC solder alloys can be improved byreducing Ag content in the alloy, thus reducing theformation of primary Ag3Sn. This enhancement in impactresistance though, often comes at the price of decreasedthermomechanical fatigue resistance due to decreased phase

fractions of small, dispersed IMC phases that can act asdislocation blockers within the solder matrix. The ideal Sn-based solder microstructure will need to be optimized forthese two properties. Given that current Sn-based soldersoften contain only one unique b-Sn nucleation site within asolder joint microstructure, a shift in grain refinement of b-Sn to include 4-5 unique b-Sn nucleation sites within asingle solder joint may mitigate the reductions in fatigue lifefrom decreased primary IMC presence by increased fatiguelife via grain refinement. Future experimental research onsolute additions and computational modeling of nucleantcluster formations within the Sn-alloy liquids should beperformed to determine the viability of such approaches inSn-based solders.

3. High Temperature Interconnect Technologies

Although the current RoHS exemption for high-Pb high-temperature solders is being reviewed by the EuropeanCommission with likely renewal through 2021, research inhigh-temperature Pb-free alternatives has intensified. This isdriven by both the need to find acceptable replacements forcurrent uses of high-Pb solders and the demand for evenhigher temperature interconnect (HTI) technologies forelectronic systems in increasingly harsh environments.The most widely used commercial Pb-free solder for hightemperature applications is the Au-20Sn wt.% eutectic alloy(Te = 278 �C).[58-63] However, due to the high price of Au,Au-Sn alloys have been used primarily in high reliabilityapplications including step-soldering in RF packages,hermetic sealing, aerospace, military, and medical electron-ics.

At the same time that alternatives to high-Pb, hightemperature solders are being sought, the operating temper-atures of integrated circuits are increasing and the environ-ments in which they are being used are becoming moreextreme. The trend to more compact auto engine compart-ments is placing control systems closer to the engines andactuators, and increasing exposure of vulnerable integratedcircuitry to higher temperatures, vibration, and thus, condi-tions for fatigue and failure. Semiconductors based on GaN,III-Vs, SiC, and diamond can operate at significantly highertemperatures than Si, whose maximum operating tempera-ture is approximately 200 �C. This creates the need for new,RoHS-compliant bonding solutions for die-attach, insulated-gate bipolar transistor (IGBT) attach, and power electronicsto replace high-Pb solders, Pb-free solders, and thermalinterface materials used today.

A wide range of non-solder alternatives have beenresearched over the last ten years resulting in a range ofcommercial products being currently offered. These tech-nologies include sintering of Ag and Cu nanoparticles, core-shell nanoparticles, mixtures of coarser Cu with Agnanoparticles, Ag-filled adhesives, Bi-Ag-X alloys, transientliquid phase sintering, and foil-based transient liquid phasebonding.[64-70] For an HTI to be an acceptable substitute forhigh-Pb solders in a given application, the HTI must:

Table I Liquidus slopes (mL, assumed linear), parti-tion coefficients (k), maximum concentrations of solutesin the Sn-liquid, and corresponding solute GRF values

Sn + XLiquidusslope (mL)

Partitioncoefficient

(k)

Maximumconcentration

(wt.%) GRF

Al �6 0.5 0.5 1

Ag �3 0.02 3.7 11

Bi �2 0.2 53 86

Co �7 0 0.02 0.1

Cu �7 0 0.9 6

In �2 0.3 4.0 6

Mn �9 0 0.02 0.2

Ni �6 0.04 0.04 0.2

Zn �4 0.05 7.8 29

A high value of GRF is desired to most effectively promote constitutional

undercooling. (Data calculated via ThermoCalc V.3 TCSLD3.[57])


1. Perform within an acceptable range of the electrical,thermal, and mechanical requirements for the applica-tion.

2. Form at temperatures less than or equal to commercialhigh-Pb solders it is replacing (<300 �C).

3. Perform reliably for typical operation and use condi-tions, including during thermal cycling to low andhigh temperatures depending on the operating ranges,which can reach 400 �C for some applications.

4. Exhibit acceptable mechanical, thermal, and electricalperformance after long-term aging and use.

Among the alternative strategies for forming HTIs,approaches using either a liquid or a solid transient phaseto meet these criteria are being widely pursued withcommercial products currently available based on severaldifferent compositions. In the case of a liquid transientphase, a low melting point liquid reacts with a highermelting point solid to solidify isothermally by interdiffusionand/or IMC formation. Two classic examples of isothermalsolidification by Rahman et al. are (1) a Cu-Ni liquid incontact with pure Ni forms a Cu-Ni solid solution byinterdiffusion with no remaining liquid phase and (2) thereaction of liquid Sn with Cu forms two IMCs on solid Cu,and if annealed long enough to reach equilibrium, forms asingle IMC on Cu. Both processes are based on isothermalsolidification by interdiffusion; the key difference betweenthe two is that the liquid is eliminated by the formation of anintermetallic phase in the Cu-Sn example versus interdiffu-sion to form a Ni-Cu solid solution in the Cu-Ni exam-ple.[71] It is interesting to note that similar approaches can beused with a transient solid phase to raise the meltingtemperature of the remaining phases in the solder intercon-nect. For example, Zhu et al. recently developed a transientsolid reaction system by first heating a solid tri-layerinterconnect, Ni/eutectic Au-20Sn wt.%/Ni, to above theeutectic temperature (278 �C), then cooling it to 240 �C tosolidify the liquid to form AuSn and Au5Sn (1’ phase), andfinally transforming it into a Ni-Ni3Sn4-Au5Sn-Ni3Sn4-Nimultilayer structure by annealing the tri-layer for 100 h at240 �C. The resulting 1’ phase shows a region of significantsolid solubility, with the IMC composition shifting to higherAu concentrations and higher temperature stability asNi3Sn4 continues to form.[63]

Transient phase processes in Sn-based systems have beenreferred to by various names and acronyms by differentresearch groups. The terms ‘‘transient liquid phase sinter-ing’’ (TLPS), ‘‘liquid phase diffusion bonding’’ (LPDB),‘‘transient liquid phase bonding’’ (TLPB), and solid-liquidinterdiffusion (SLID) have all been used to describebonding of two substrates via isothermal solidificationreactions between a low melting temperature phase (LTP)and the high melting temperature (HTP) substrates.[63,67,69]

In this paper, the term LPDB will be used, although all canbe used interchangeably. During processing when thetemperature is heated higher than the solidus of the LTP,liquid forms and wets the HTP and the substrate surfaces.Typically the LTP is heated above its liquidus so that theLTP is completely molten. Isothermal solidification occursby interdiffusion such that no liquid remains. The role of

added HTP particles is to increase the surface area forinterdiffusion and reaction, to reduce the time needed forsufficient reaction, rather than being limited by the surfacearea of the substrates. The underlying principle is the samewith or without HTP particles added to the LTP. In thispaper we focus on two approaches to HTI alloy and systemdesign for LPDB systems for different applications. The firstapproach illustrates how metastability can be exploited inthe Cu-Ni-Sn ternary to create a pseudo-binary LPDBsystem using a Sn-Cu-Ni LTP alloy and Cu-10Ni wt.%binary solid solution HTP. The second approach presents athermodynamic framework to bound alloy candidates forternary and higher order LPDB systems as a function ofprocessing and operating temperatures.

3.1 Liquid Phase Diffusion Bonding in the Cu-Ni-SnSystem

Liquid phase diffusion bonding reactions in the Sn-Cu-Ni system were examined by McCluskey, Greve, andMoeini with a Sn-3.5Ag LTP in combination with Cu, Ni, ormixture of Cu and Ni powders with Ni substrates as theHTP substrates.[67,68,72] Samples were processed at a peaktemperature of 300 �C in an inert atmosphere for approx-imately 30 minutes using 0.2 MPa of applied pressure toreduce void formation. The resulting microstructures for Ni-Sn and (Cu, Ni)-Sn contained few voids; pockets of Sn incross-sections of the bonds indicated incomplete solidifica-tion. Despite the presence of unreacted Sn, shear tests of Agmetallized Si dies bonded to Ni substrates using thistechnique indicated that the bonds were able to withstand 10MPa up to 435 �C using mixed Cu and Ni particles as theHTP and up to 600 �C (test setup limit) with Ni HTPparticles.[73] These mechanical property results are consis-tent with the melting temperatures of the expected inter-metallics in the systems: approximately 415 �C for Cu6Sn5and 798 �C for Ni3Sn4. These results also demonstrate thatthe bonds may provide adequate performance at elevatedtemperatures even in the presence of isolated pockets ofresidual liquid.

The Cu-Ni-Sn ternary system has advantages over thebinary Sn-Cu system. Nickel, even in trace amounts, issoluble in Cu6Sn5 and suppresses the IMC’s allotropic phasechange upon cooling, i.e., stabilizing the high temperaturehexagonal (Cu, Ni)6Sn5.

[37] However, as reported byMcCluskey, when both Cu and Ni HTP particles are used,the system has competing reactions between the twobinaries: with the formation of Cu3Sn and Cu6Sn5 betweenCu and Sn, and the formation of Ni3Sn4 between Ni and Sn.If the Cu6Sn5 forms quickly enough such that little Ni fromthe liquid is available to be incorporated into the IMC, thelow temperature, more brittle monoclinic Cu6Sn5 can form.Furthermore, as seen in reactions in BGAs in which onesubstrate is Cu and one is Ni, the two intermetallics incontact with the liquid can change in both composition andphase over time since they are not in thermodynamicequilibrium.

Choquette and Anderson avoided these issues by creatingLPDB systems with Cu-10Ni wt.% HTP alloy powders andSn-0.7Cu-0.05Ni LTP powder (Nihon Superior, SN100C),


leading to the formation of (Cu, Ni)6Sn5 in the processedjoints. The microstructures formed by this powder pastereflowed with 30s above liquidus (227 �C) to a peaktemperature of 250 �C can be seen in Fig. 5. Liquid-phasediffusion bonding occurred between the tin alloy (white)and the solid Cu-Ni powder (dark gray), with reactionformation of the (Cu, Ni)6Sn5 phase (light gray). Noformation of (Cu, Ni)3Sn was observed. The absence of (Cu,

Ni)3Sn has been reported previously for similar diffusioncouples annealed at 240 �C by Vuorinen et al. and at 200 �Cby Baheti et al.[74,75] Baheti et al. presented evidence that(Cu, Ni)3Sn is not thermodynamically stable at 200 �C forNi concentrations equal to or greater than 7.5 wt.% Ni, andsuggested that the absence of the thermodynamicallystable ternary intermetallic was due to difficulty in nucle-ation of the phase.[75] However, at 240 �C, Vuorinen et al.argued that the absence of the (Cu, Ni)3Sn was kinetic inorigin, i.e., that (Cu, Ni)3Sn was thermodynamically stable,but did not form due to rapid growth of (Cu, Ni)6Sn5.

[74]

This latter interpretation is supported by phase equilibriumexperiments by Lin et al. that showed (Cu, Ni)3Sn inequilibrium with Cu-Ni alloys at 240 �C.[76] The mostimportant feature of this system is, therefore, that it can betreated as two pseudo-binaries. This can be understood bycomparing the 240 �C isothermal section of the phasediagram (Fig. 6a) with what we refer to as a ‘‘reactiondiagram’’ (Fig. 6b). This isothermal reaction diagramschematically shows the phases observed during the shortannealing times required for LPDB systems. Many binaryand ternary phases present in the equilibrium phase diagramare missing in the diffusion couples as discussed above. TheCu-Ni HTP alloys in contact with Sn form either (Cu,Ni)6Sn5 or (Ni, Cu)3Sn4 depending on the HTP alloycomposition, and at low Ni concentrations (Cu, Ni)3Sn isalso observed. This diagram helps to visualize the resultingphases forming from diffusion couples in various HTP andLTP composition regimes.

Recent experiments testing several volume ratios ofLTP:HTP (Sn alloy powder:Cu-10Ni wt.% powder) showedformation of large voids and flux trapping at a ratio of 25

Fig. 5 LPDB microstructure of SN100C LTP and Cu-10 wt.%Ni HTP particles (dark grey). (Cu, Ni)6Sn5 intermetallic com-pound (light grey), and residual Sn (white) (Previously unpub-lished micrograph provided by the author)

Fig. 6 (a) Calculated metastable equilibria in the Sn-Cu-Ni systems at 240 �C. The possible s-phase is indicated and dotted lines in theCu-Ni side of the diagram beyond the formation of (Cu, Ni)6Sn5 and (Ni, Cu)3Sn4 reflect the uncertainties in the phase equilibria. Plotreproduced from Ref 74. (b) Schematic reaction diagram represents the phases observed in diffusion couples composed of Cu-Ni HTPand Sn at around 240 �C. The gold circles indicate the final phases remaining after complete consumption of the Sn for a Cu-10 wt.%NiHTP alloy (Color figure available online)


vol.%:75 vol.% (LTP:HTP). This issue was resolved whenthe ratio of LTP:HTP was increased to 75 vol. %:25 vol.%(LTP:HTP). This increased fraction of liquid phase allowedfor better wetting, greater pore filling, significant rearrange-ment within the LTP/HTP mixture before IMC formation,and the flux to be expelled, thus reducing porosity. A typicalmicrostructure resulting from the 75 vol. % LTP:25 vol.%HTP ratio is shown in Fig. 7, after 30s reflow aboveliquidus (227 �C) with a peak reflow temperature of 250 �C.Approximately 10 lm of (Cu, Ni)6Sn5 was observed to haveformed and significant residual Sn-liquid was left surround-ing the HTP particles. For a viable LPDB system, there mustbe significant ‘‘bridge’’ formation (IMC growth) betweenthe (Cu, Ni)6Sn5 coated HTP particles and percolation of theIMC to form a rigid structure. In Fig. 7 there is littleevidence that this has occurred. As noted above, the resultsof Greve et al. indicated that complete consumption of theSn is not required, however, the acceptable levels of residualliquid as a function of starting microstructure are notknown.[73] There are at least two minimum conditions forbridge formation. First, there must be enough HTP alloy toconsume all the Sn; for the formation of Cu6Sn5 between Cuand Sn, the critical ratio is 35 vol.%:65 vol.% (Cu:Sn).Therefore, the amount of HTP will need to be increased forthe reaction to continue. The second is that the (Cu,Ni)6Sn5-coated HTP particles must be touching such that theresidual liquid is isolated at the HTP particle junctions.Whether this will occur during processing for a given ratioof LTP:HTP will depend on a number of factors, includingthe reaction rate and the starting HTP particle size. It isinteresting to note that Vuorinen et al. reported that thegrowth rate of (Cu, Ni)6Sn5 at 240 �C increased as afunction of Ni in the Cu-Ni alloy, reaching a maximum atapproximately 10 at.% Ni.[74] In their experiments with Cu-

10 at.%Ni in contact with Sn, the (Cu, Ni)6Sn5 thicknesswas approximately 10 lm after 10 min at 240 �C, incontrast to the 10 lm after 30s at 250 �C seen here byChoquette and Anderson.[69] Optimization of the Cu-Ni-SnLPDB system will require quantifying the effects ofLTP:HTP ratio, HTP particle size, packing, and processingtemperature on liquid consumption rate, the location of theresidual liquid, and the resulting mechanical properties ofthe LPDB structure. Practical processing and compatibilityconsiderations will dictate the maximum acceptable process-ing temperatures and times, and therefore, under whatconditions the Cu-Ni-Sn LPDB system can be used.

3.2 Thermodynamic Framework for Identifying NewTernary LPDB Systems

The relationship between the equilibrium phases in Sn-based ternary phase diagrams and the specific interdiffusionpaths for soldering and bond formation have been studiedextensively for a range of systems. For example, Chen et al.,determined both the phase diagrams and the diffusion pathsfor a range of Sn-In alloys reacting with Ag.[77] Thesestudies and the relationships they have revealed can beleveraged to identify new LPDB formulations and process-ing paths for multi-component systems. In particular, wesuggest that design criteria can be defined and used as ascreening tool to identify potential LTP and HTP compo-sitions based on examination of ternary phase diagrams,either calculated or experimental. These criteria describedbelow are based on (1) melting temperature of the LTP, (2)formation of IMCs that consume components of the liquidphase, (3) three phase equilibrium between the terminalliquid, the IMC, and any third phase that must precipitate forthe liquid fraction to decrease, and (4) the temperature atwhich reactions occur in the solidified structure as it isheated.

The binary Cu-Sn LPDB system provides a simplestarting point before proceeding to discuss three-componentsystems. The binary Cu-Sn phase diagram shown in Fig. 8illustrates several of the key thermodynamic characteristicsnecessary for LPDB. The first characteristic is that process-ing must occur above the melting point of the LTP, which inthe case of pure Sn is 232 �C. The second characteristic isthat the resulting intermetallic phases from a given HTP-LTP combination incorporate Sn, the primary component inthe liquid. Consider the interface between a small volume ofliquid Sn in contact with a relatively large volume of solidCu at 250 �C. The blue, double-arrow line in Fig. 8identifies the 250 �C isotherm on the Cu-Sn phase diagram.Initially, Cu dissolves into the liquid Sn. (It is assumed thatthe liquid is well mixed.) Once the Sn becomes saturatedwith Cu, Cu6Sn5 nucleates. The Cu6Sn5 typically develops ascalloped interface, and a Cu3Sn layer forms between the Cuand Cu6Sn5. There are now 3 interfaces: Cu-Cu3Sn, Cu3Sn-Cu6Sn5, and Cu6Sn5-Sn. The evolution of phases at the Cu-Sn interface follows the line across the isotherm, indicatedin Fig. 8.

Ternary elements may be added to the LTP for severalpurposes. An alloying component may be used to modifythe IMC phases in equilibrium with the liquid LTP.

Fig. 7 LPDB microstructure of 25 vol. % Cu-10 wt.% Ni HTPparticles (black) and 75 vol.% SN100C LTP after reflow process-ing for 30s above the liquidus (227 �C) with a peak temperatureof 250 �C (Previously unpublished micrograph provided by theauthor)


Alternative IMC phases may be favored based on mechan-ical properties or melting temperature. A three-componentsystem also results in the regions of three-phase equilibrium.Solute rich alloys can be therefore leveraged to reduceprocessing time by precipitation of a new phase in the liquidas the IMC forms, i.e., solidification occurs both by IMCformation and precipitation. The precipitation of a newphase is not itself diffusion limited, but it is determined bythe rate of IMC formation. A liquid composition in three-phase equilibrium with the dominant IMC phase and theprecipitate phase is necessary to gain this benefit of twoconcurrent forms of isothermal solidification. The thirdcharacteristic is that the terminal liquid must be inequilibrium with both the IMC and the third precipitatingphase in order for the liquid to be consumed. The fourthcharacteristic is determined by the use temperature: theprecipitating phase must have a melting point higher thanthe intended use temperature, and the precipitating phasemust not react to form a liquid at the intended usetemperature.

In the discussion that follows, these four thermodynamiccharacteristics are discussed for three Sn-Bi ternaries: Ag-Bi-Sn, Cu-Bi-Sn, and Ni-Bi-Sn, and three Sn-In ternaries:Ag-In-Sn, Cu-In-Sn, and Ni-In-Sn. Phase diagrams werecalculated using Thermo-Calc (version 3) with the soldersolution database version 1.[57] Isothermal sections at 1 �Cintervals were generated and compared with experimentalresults from the literature, such as Ref 78.

3.3 Potential LPDB Systems

3.3.1 Sn-Ag, Sn-Cu, or Sn-Ni. Typical substrate andHTP powder compositions are composed of Ag, Cu, or Ni,and form the basis of the starting binaries. The elements Agand Cu form eutectics with Sn at 220 and 227 �C,respectively. Nickel does not significantly decrease themelting temperature of Sn, and therefore does not affect theLTP melting temperature, but it may be used to modify theinitial equilibrium phase with Cu, as in the Cu-Ni-Snsystem. Because all three HTPs form IMC phases with Sn,effective solute compositions are limited to the liquiduscomposition in equilibrium with the IMC at the isothermalsolidification temperature. Hyper-eutectic LTP compositionswill result in primary IMC in the liquid that remains solidduring processing, and may be of benefit to reduce theamount of IMC formation necessary to isothermally solidifyand form interconnects.

3.3.2 Bi-Sn LTP with Ag, Cu, or Ni. The Sn-Bi eutecticreaction occurs near 138 �C and 57 wt.% Bi (43 at.% Bi).The element Bi forms simple binary eutectics with Ag(Te = 262.5 �C) and Cu (Te = 270.6 �C), with no IMCs.The Ni-Bi phase diagram contains two IMCs, NiBi andNiBi3, and has a eutectic reaction (Te = 271 �C), however,the NiBi3 IMC has been observed only in long-termannealing experiments after reaction between Ni and Sn-Bi alloys with Bi content in the alloy >96.5 at.% Bi.[79]

Given the deep Sn-Bi eutectic, the maximum effectiveaddition of Bi is limited only by the compositions that are

Fig. 8 Cu-Sn phase diagram generated using ThermoCalc with database TCSLD1.[57] The gold circles indicate the phases observedwhen the liquid phase is consumed in typical Cu-Sn LPDB reactions. Further annealing may lead to the disappearance of at least one ofthe two phases depending on the interdiffusion kinetics (Color figure available online)


liquid at the targeted processing temperature. In fact, theamount of Bi solute may be leveraged to minimize theamount of Sn that must be consumed by IMC formation forisothermal solidification. The ternary Ag-Bi-Sn (Ag HTPsubstrate) will form a single IMC Ag3Sn, with the rejectedBi phase precipitating from the terminal liquid as the IMCgrows. The isothermal sections from the phase diagram andthe corresponding reaction diagrams contain the samephases, with the exception of the 1-Ag phase. The ternaryNi-Sn-Bi is also compatible with LPDB, since Ni3Sn4 is theonly IMC that forms in diffusion couples with Sn-Bi alloys,and it is in equilibrium with the rejected Bi phase. For Bi-Ni-Sn, the isothermal sections of the phase diagram and thecorresponding reaction diagrams are similar, with theexception of the Ni-Bi IMCs.

The ternary Cu-Sn-Bi is both simpler, and yet, morecomplicated than the Ag-Bi-Sn and Bi-Ni-Sn systems,primarily due to the two equilibrium Cu-Sn IMCs (Cu6Sn5and Cu3Sn) that also form in the diffusion couples. Since Bidoes not react with Cu, it must precipitate as a solid in orderfor isothermal solidification to proceed. Both Cu-Sn IMCphases are in equilibrium with Bi below 200 �C, but onlyCu3Sn is in equilibrium with Bi above 200 �C. The Cu-Sn-Bi ternary system exhibits the following Class II reaction atapproximately 200 �C: Cu3Sn + L fi Cu6Sn5 + Bi.[80]

Isothermal sections above and below 200 �C are shown inFig. 9. Since all the equilibrium phases are observed indiffusion couples, isothermal sections can be used asreaction diagrams. At 170 �C, reactions for all Sn-Bi LTPalloy compositions to the left of the terminal liquidcomposition in contact with Cu lead to the formation of

Cu6Sn5 in contact with the liquid. As the IMC grows, theliquid composition shifts until it reaches the terminal liquidcomposition (blue star, Fig. 9a) for the three-phase equilib-rium triangle for Cu6Sn5 and solid Bi. As the IMC continuesto grow, the amount of liquid also decreases by precipitationof the Bi, until all liquid is consumed. Additionally, there isthen also the formation of the Cu3Sn phase between Cu andCu6Sn5. The final phases in the LPDB structure are thoseindicated by the gold circles (Fig. 9a). In contrast, attemperatures above 200 �C (260 �C isothermal shown inFig. 9b), Sn-rich Sn-Bi liquids reach a terminal liquidcomposition (blue star, Fig. 9b) in equilibrium with Cu6Sn5and Cu3Sn, not with solid Bi. Therefore, for Sn-rich Sn-Bialloys, the growth of Cu6Sn5 stops at this terminal liquidcomposition, and, hence, isothermal solidification proceedsuntil the Cu6Sn5 is completely consumed by growth ofCu3Sn (gold circles indicate final phases, Fig. 9b). Isother-mal solidification is possible as low as the eutectictemperature for Sn-rich Sn-Bi alloys, but reaction of Cu6Sn5and rejected Bi to form liquid limits processing andoperating temperatures to below 200 �C. In comparison,for Bi-rich Sn-Bi alloy compositions above 200 �C, Cu3Snformation occurs in equilibrium with the Bi-rich liquid,shifting to higher Bi concentrations as Cu3Sn growthproceeds. With the terminal liquid composition indicated(black star, Fig. 9b), the reaction can go to completionavoiding the limitations posed by the equilibrium inversionat Sn-rich concentrations, with the final phases in the LPDBstructure indicated by the black circles (Fig. 9b) Thelimiting factor for any of these ternaries with Sn-Bi is themelting temperature of Bi, 271 �C. The ternary Sn-Bi LTPs

Fig. 9 Isothermal sections of the Cu-Sn-Bi phase diagram generated using Thermo-Calc V.3 with database TCSLD1.[57] (a) At 170 �Call initial LTP liquid compositions can be isothermally solidified with the terminal liquid composition (blue star) as marked. The goldcircles indicate the phases observed when the liquid phase is consumed in typical LPDB reactions. (b) At 260 �C, Sn-rich LTP composi-tions end up at the terminal liquid composition (blue star) which leads to a stable liquid as long as Cu6Sn5 is in contact with the liquid.During reaction, Bi-rich LTP compositions shift to the terminal liquid composition (black star) and further reaction leads to the phasesindicated by the black circles (Color figure available online)


with Ag, Cu, or Ni are thus recommended for furtherinvestigation.

3.3.3 In-Sn LTP with Ag, Cu, or Ni. In the binary, Snand In react to form binary IMCs with extensive solidsolubility; the Sn-In binary eutectic reaction is between twointermetallics and occurs near 120 �C and 50.9 wt.% In(51.7 at.% In). Both In and Sn form IMCs with Ag, Cu, andNi. There have been a few observations of ternarycompounds, but there are regions of complete solidsolubility across the ternaries and more limited solubilityof the third element in some binary IMCs. The ternary Ag-Sn-In is minimally compatible with LPDB, because In-Snalloys react with Ag to form Ag-In IMCs with limited solidsolubility of Sn. With low temperature processing, this canlead to the precipitation of a Sn-In IMC that melts at222 �C, and therefore, does not offer any advantages overconventional Pb-free solder alloys. In contrast, the ternaryCu-Sn-In is a promising LPDB system because the Cu6(Sn,In)5 phase in equilibrium with near-eutectic liquid compo-sitions incorporates both components of the LTP, as seen inthe calculated Cn-Sn-In isothermal section of the ternaryphase diagram in Fig.10, and has been shown to solidifyisothermally by Sasangka et al.[81] The ternary Ni-Sn-Inexhibits similar characteristics to Cu-Sn-In, but the solubil-

ity of In in Ni3Sn4 is more limited. As with the Ag-In-Snsystem, alloys of In-rich Ni-Sn-In are likely to formundesirable low-melting temperature In-Sn IMCs ratherthan Ni3(Sn, In)4. The ternaries of In-Sn LTPs with Cu andNi exhibit promising characteristics over specific composi-tion and temperature ranges and are, therefore, recom-mended for further investigation, particularly the Cu-In-Snsystem.

3.3.4 Calculated Phase Diagrams and the Role ofKinetics. As demonstrated above, thermody-

namic criteria for LPDB can be used to identify compositionand temperature ranges that exist in promising ternarysystems and warrant further study. The calculated isothermalsections and solidification paths can then be used as astarting point for understanding IMC formation andmicrostructural evolution. However, thermodynamics isonly one aspect of selecting a formulation. The kinetics ofIMC phase formation may change for a number of reasonswith the addition of ternary components. As noted above,Baheti et al. found that increasing the Ni concentration inCu-Ni alloys (£5 at.% Ni) decreased the growth rate of (Cu,Ni)3Sn until at Cu-5 at.% Ni, (Cu, Ni)3Sn was not presenteven though it was thermodynamically stable. This wasattributed to the increase in the growth rate of (Cu, Ni)6Sn5

Fig. 10 Isothermal section of the Cu-Sn-In phase diagram at 170 �C generated using Thermo-Calc V.3 with database TCSLD1.[57] Allinitial LTP liquid compositions can possibly be isothermally solidified forming a Cu6(Sn, In)5 ternary solid solution. Experimental mea-surements of the specific reaction paths for different LTP compositions and the relative diffusivities of Sn and In in the IMCs still mustbe evaluated to determine what ranges can be used (Color figure available online)


with increasing Ni concentrations, due to both the change inthe relative diffusivities of Sn and Cu in the (Cu, Ni)6Sn5phase and the increasing contribution of grain boundarydiffusion to growth as the (Cu, Ni)6Sn5 grain size decreasedwith increasing Ni concentration.[75] Stable phases atoperating temperatures must be considered as well. Forinstance, Chen et al. found that Sn-20In wt.% reacts with Agto form a ternary IMC phase over a wide range ofcompositions at 250 �C, but aging at 125 �C caused aAgIn2 phase to form. Chen et al. also observed variation inthe growth kinetics of the ternary Ag-Sn-In phase dependingon the composition of the Sn-In phase.[77] Therefore,identifying the diffusion pathway and kinetics for eachpotential LPDB system of interest will be highly relevantand vital, in addition to the thermodynamic characteristicsof the LPDB system, to the success and implementation offuture LPDB systems for high temperature solder intercon-nects.

4. Conclusions

The need for new Pb-free solders and high temperatureinterconnects has continued long after the RoHS-driven, Pb-free solder transition due to demand for both high reliabilityand good performance in increasingly harsh environments.Alloy thermodynamics and phase diagrams, as representedby the work of Jack Smith, are foundational in ourcontinuing efforts to design Pb-free interconnect solutionswith improved performance and for high temperatureinterfaces. This paper highlights some recent advances inmicrostructure design for Pb-free interconnect systems,focusing on existing challenges in the nucleation of b-Sn inSn-based solder alloys and on alloy design to createisothermally solidified interconnects using transient liquidphases.

The progress made since Swenson’s 2007 review of thechallenges in b-Sn nucleation can be understood in thecontext of Swenson’s three approaches for improving b-Snnucleation: decreasing, or eliminating, Ag from Sn-basedalloy design to advantage non-equilibrium phase formation,the addition of nucleating heterogeneities to promote b-Sngrain nucleation, and the use of rapid solidification todecrease, or eliminate, primary IMC formation. Despitethese advances made based on Swenson’s suggestions, a Sn-based alloy has yet to be identified that can holisticallycontrol solder IMC formation, b-Sn undercooling, andrefinement of b-Sn grain size (with multiple non-twinnedorientations). Thus, future work was proposed via a multi-faceted solidification approach combining the benefits ofconstitutional undercooling with the addition of solute andthe effects of heterogeneous nucleating particles to ulti-mately promote b-Sn grain refinement in Sn-based solderalloys. Using Al alloys as the model, growth restrictionfactors (GRFs) were used to define potential solute additionsto pure Sn that promote constitutional undercooling, andthus enhanced nucleation of the b-Sn phase. One suchsuggestion was the addition of Zn in the range of 3-8 wt.%

(GRFs = 11-29), utilized in tandem with uniformly dis-persed heterogeneous nucleant particles, such as XSn4IMCs, which could hold promise for providing both theconstitutional undercooling and the potent nucleation sitesnecessary for enhanced nucleation and ultimate grainrefinement the b-Sn phase. Such a multi-faceted approachto enhancing the nucleation of b-Sn and thus reducing theoverall b-Sn grain size in Sn-based solder alloys holds muchpromise and future research, both experimentally andcomputationally, in these areas of b-Sn solidification shouldbe pursued.

In the final section, two different approaches to inter-connect design were presented based on liquid phasediffusion bonding (LPDB) as a replacement for high-Pbsolders based on the Cu-Ni-Sn system and on a thermody-namic framework for identifying promising ternary alloysfor LPDB. Modification of the equilibrium phases by usingCu-Ni alloys as the high melting temperature phase waspursued in order to improve performance previouslydemonstrated using LPDB systems containing Cu-Sn andCu-Ni-Sn with separate Ni and Cu particles. A frameworkfor selecting alternative LPDB formulations was alsodescribed: the equilibrium phase diagrams were used as astarting point for down-selection based on four thermody-namic criteria. This framework was applied to existing datafor three Sn-Bi ternaries (Sn-Bi-Ag, Sn-Bi-Cu, and Sn-Bi-Ni) and three Sn-In ternaries (Sn-In-Ag, Sn-In-Cu, and Sn-In-Ni), and specific ternaries and composition ranges wereidentified as promising low temperature phases for LPDB. Aspecial form of metastable diagram for LPDB which werefer to as a ‘‘reaction diagram’’ was introduced as a way tounderstand the observed deviations from equilibrium phasediagrams for isothermal annealing as a function of temper-ature and time. Further development of Pb-free solder alloysand LPDB interconnect solutions will continue to buildupon the foundations of thermodynamics, to which JackSmith so significantly contributed throughout his career.

Acknowledgments

This work was supported by Ames Laboratory, the NSFCooling Technologies Research Center at Purdue University(NSF I/UCRC Grant IIP 0649702), Nihon Superior throughAmes Lab Contract No.DE-AC02-07CH11358, and gov-ernment support under and awarded by DoD, Air ForceOffice of Scientific Research, National Defense Science andEngineering Graduate (NDSEG) Fellowship, 32 CFR 168a.

References

1. J.F. Smith, Thermodynamic Properties of Binary LithiumSystems—A Review, J. Nucl. Mater., 1976, 59(2), p 158-174

2. H.M. Kandil, J.D. Greiner, and J.F. Smith, Single-CrystalElastic Constants of Yttria-Stabilized Zirconia in the Range 20to 700 �C, J. Am. Ceram. Soc., 1984, 67(5), p 341-346

3. C.M. Miller, I.E. Anderson, and J.F. Smith, AViable Tin-LeadSolder Substitute: Sn-Ag-Cu, J. Electron. Mater., 1994, 23(7),p 595-601


4. European Parliment, Directive 2012/19/EU of the EuropeanParliment and the Council on Waste Electrical and ElectronicEquipment (WEEE)—recast. 2012, pp. 38-71

5. European Parliament, Directive 2011/65/EU of the EuropeanParliment and of the council on the Restriction of the Use ofCertain Hazardous Substances in Electrical and ElectronicEquipment (RoHS)—recast. 2011, pp. 88-110

6. V. Eveloy, S. Ganesan, Y. Fukuda, J. Wu, and M.G. Pecht, AreYou Ready for Lead-Free Electronics?, IEEE Trans. Compo-nents Packag. Technol., 2005, 28(4), p 884-894

7. K.J. Puttlitz, Overview of Lead-Free Solder Issues IncludingSelection, Handbook of Lead-Free Solder Technology forMicroelectronic Assemblies, K.J. Puttlitz and K.A. Stalter, Ed.,Marcel Decker Inc, New York, 2004, p 3-4

8. D.R. Frear, Issues Related to the Implementation of Pb-freeElectronic Solders in Consumer electronics, J. Mater. Sci.Mater. Electron., 2006, 18(1-3), p 319-330

9. K.N. Tu, A.M. Gusak, and M. Li, Physics and MaterialsChallenges for Lead-Free Solders, J. Appl. Phys., 2003, 93(3),p 1335

10. I. Karakaya and W.T. Thompson, The Pb-Sn (Lead-Tin)system, J. Phase Equilibria, 1988, 9(2), p 144-152

11. K.-W. Moon, W.J. Boettinger, U.R. Kattner, F. Biancaniello,and C.A. Handwerker, Experimental and ThermodynamicAssessment of Sn-Ag-Cu Solder Alloys, J. Electron. Mater.,2000, 29(10), p 1122-1136

12. A. Ohno and T. Motegi, The Effect of Small Additives on theUndercooling of Pure Tin, J. Japan Inst. Met., 1973, 37(7), p777-780

13. I.E. Anderson, J.C. Foley, B.A. Cook, J. Harringa, R.L.Terpstra, and O. Unal, Alloying Effects in Near-Eutectic Sn-Ag-Cu Solder Alloys for Improved Microstructural Stability,J. Electron. Mater., 2001, 30(9), p 1050-1059

14. K.S. Kim, S.H. Huh, and K. Suganuma, Effects of FourthAlloying Additive on Microstructures and Tensile Propertiesof Sn-Ag-Cu Alloy and Joints with Cu,Microelectron. Reliab.,2003, 43(2), p 259-267

15. I.E. Anderson, Development of Sn-Ag-Cu and Sn-Ag-Cu-XAlloys for Pb-free Electronic Solder Applications, J. Mater.Sci. Mater. Electron., 2007, 18(1), p 55-76

16. L.-W. Lin, J.-M. Song, Y.-S. Lai, Y.-T. Chiu, N.-C. Lee, and J.-Y.Uan, AlloyingModification of Sn-Ag-Cu Solders byManganeseand Titanium,Microelectron. Reliab., 2009, 49(3), p 235-241

17. I.E. Anderson, J.W. Walleser, J.L. Harringa, F. Laabs, and A.Kracher, Nucleation Control and Thermal Aging Resistance ofNear-Eutectic Sn-Ag-Cu-X Solder Joints by Alloy Design, J.Electron. Mater., 2009, 38(12), p 2770-2779

18. A.J. Boesenberg, I.E. Anderson, and J.L. Harringa, Develop-ment of Sn-Ag-Cu-X Solders for Electronic Assembly byMicro-Alloying with Al, J. Electron. Mater., 2012, 41(7), p1868-1881

19. K. Sweatman, T. Nishimura, S. D. McDonald, M. Whitewick,and K. Nogita, Grain Refinement for Improved Lead-FreeSolder Joint Reliability, SMT Magazine, pp. 30-40, 2014

20. K.N. Reeve, I.E. Anderson, and C.A. Handwerker, Nucleationand Growth of Cu-Al Intermetallics in Al-Modified Sn-Cu andSn-Ag-Cu Lead-Free Solder Alloys, J. Electron. Mater., 2015,44(3), p 842-866

21. D. Swenson, The Effects of Suppressed Beta Tin Nucleationon the Microstructural Evolution of Lead-Free Solder Joints, J.Electron. Mater., 2007, 18(1-3), p 39-54

22. J.H. Perepezko, Nucleation in Undercooled Liquids, Mater.Sci. Eng., 1984, 65(1), p 125-135

23. L.P. Lehman, S.N. Athavale, T.Z. Fullem, A.C. Giamis, R.K.Kinyanjui, M. Lowenstein, K. Mather, R. Patel, D. Rae, J.

Wang, Y. Xing, L. Zavalij, P. Borgesen, and E.J. Cotts, Growthof Sn and Intermetallic Compounds in Sn-Ag-Cu Solder, J.Electron. Mater., 2004, 33(12), p 1429-1439

24. L.P. Lehman, Y. Xing, T.R. Bieler, and E.J. Cotts, Cyclic TwinNucleation in Tin-based Solder Alloys, Acta Mater., 2010,58(10), p 3546-3556

25. J.-M. Song, J.-J. Lin, C.-F. Huang, and H.-Y. Chuang,Crystallization, Morphology and Distribution of Ag3Sn inSn-Ag-Cu Alloys and their Influence on the Vibration FractureProperties, Mater. Sci. Eng. A, 2007, 466(1), p 9-17

26. T.R. Bieler, L.P. Lehman, T. Kirkpatrick, E.J. Cotts, and B.Nandagopal, Influence of Sn Grain Size and Orientation on theThermomechanical Response and Reliability of Pb-free SolderJoints, IEEE Trans. Components Packag. Technol., 2008,31(2), p 370-381

27. A.U. Telang and T.R. Bieler, Characterization of Microstruc-ture and Crystal Orientation of the Tin Phase in Single ShearLap Sn-3.5Ag Solder Joint Specimens, Scr. Mater., 2005,52(10), p 1027-1031

28. K.L. Buckmaster, J.J. Dziedzic, M.A. Masters, B.D. Poquette,G.W. Tormoen, D. Swenson, D.W. Henderson, T. Gosselin,S.K. Kang, D.Y. Shih, and K.J. Puttlitz, The Effects of MinorZn Alloying Additions on the Solidification Behavior andMicrostructure of Near-Eutectic Sn-Ag-Cu Solder Joints, inPresented at the ‘‘Pb-Free and Pb-Bearing Solders: Solidifi-cation’’ Symposium at the Materials Science & Technology(MS&T) Conference 2003 Fall Meeting

29. S.K. Kang, D.Y. Shih, D. Leonard, D.W. Henderson, T.Gosselin, S. Cho, J. Yu, and W.K. Choi, Controlling Ag3SnPlate Formation in Near-Ternary-Eutectic Sn-Ag-Cu Solder byMinor Zn Alloying, J. Met., 2004, 56(6), p 34-38

30. K.S. Kim, S.H. Huh, and K. Suganuma, Effects of CoolingSpeed on Microstructure and Tensile Properties of Sn-Ag-CuAlloys, Mater. Sci. Eng. A, 2002, 333(1-2), p 106-114

31. D.W. Henderson, T. Gosselin, A. Sarkhel, S.K. Kang, W.Choi, D. Shih, C. Goldsmith, and K.J. Puttlitz, Ag3Sn PlateFormation in the Solidification of Near Ternary Eutectic Sn-Ag-Cu Alloys, J. Mater. Res., 2002, 17(11), p 2775-2778

32. D. Suh, D.W. Kim, P. Liu, H. Kim, J.A. Weninger, C.M.Kumar, A. Prasad, B.W. Grimsley, and H.B. Tejada, Effects ofAg Content on Fracture Resistance of Sn-Ag-Cu Lead-FreeSolders under High-Strain Rate Conditions,Mater. Sci. Eng. A,2007, 460-461(July), p 595-603

33. H. Tsukamoto, T. Nishimura, S. Suenaga, S.D. Mcdonald, K.W.Sweatman, and K. Nogita, The Influence of Solder Compositionon the Impact Strength of Lead-Free Solder Ball Grid ArrayJoints, Microelectron. Reliab., 2011, 51(3), p 657-667

34. C.E. Ho, R.Y. Tsai, Y.L. Lin, and C.R. Kao, Effect of CuConcentration on the Reactions between Sn-Ag-Cu Soldersand Ni, J. Electron. Mater., 2002, 31(6), p 584-590

35. C.W. Chang, S.C. Yang, C.-T. Tu, and C.R. Kao, Cross-Interaction between Ni and Cu across Sn Layers with DifferentThickness, J. Electron. Mater., 2007, 36(11), p 1455-1461

36. W.M. Chen, T.L. Yang, C.K. Chung, and C.R. Kao, TheOrientation Relationship betweenNi and Cu6Sn5 Formed duringthe Soldering Reaction, Scr. Mater., 2011, 65(4), p 331-334

37. K. Nogita and T. Nishimura, Nickel-Stabilized Hexagonal (Cu,Ni)6Sn5 in Sn-Cu-Ni Lead-free Solder Alloys, Scr. Mater.,2008, 59(2), p 191-194

38. K. Nogita, C.M. Gourlay, and T. Nishimura, Cracking andPhase Stability in Reaction Layers between Sn-Cu-Ni Soldersand Cu Substrates, J. Mater., 2009, 61(6), p 45-51

39. K. Nogita, Stabilisation of Cu6Sn5 by Ni in Sn-0.7Cu-0.05NiLead-free Solder Alloys, Intermetallics, 2010, 18(1), p 145-149


40. K. Nogita, D. Mu, S.D. Mcdonald, J. Read, and Y.Q. Wu,Effect of Ni on Phase Stability and Thermal Expansion ofCu(6-x)Ni(x)Sn5 (X = 0, 0.5, 1, 1.5 and 2), Intermetallics,2012, 26(July), p 78-85

41. J.W. Xian, S.A. Belyakov, T.B. Britton, and C.M. Gourlay,Heterogeneous Nucleation of Cu6Sn5 in Sn-Cu-Al Solders, J.Alloys Compd., 2015, 619(January), p 345-355

42. J.W. Xian, S.A. Belyakov, and C.M. Gourlay, ControllingBulk Cu6Sn5 Nucleation in Sn0.7Cu/Cu Joints with Al Micro-alloying, J. Electron. Mater., 2015, 45(1), p 69-78

43. K. N. Reeve, S. M. Choquette, I. E. Anderson, and C. A.Handwerker, Rapid Solidification of Micro-Alloyed Sn-Cu-AlAlloys for High Reliability Lead-Free Solder: Part I.Microstructural Characterization of Rapidly Solidified Solders,Metall. Mater. Trans. A, (accepted), March 2016

44. K. N. Reeve, S. M. Choquette, I. E. Anderson, and C. A.Handwerker, Rapid Solidification of Micro-Alloyed Sn-Cu-AlAlloys for High Reliability Lead-Free Solder: Part II. Inter-metallic Coarsening Behavior of Rapidly Solidified Soldersafter Multiple Reflows, Metall. Mater. Trans. A, (underreview), March 2016

45. S.A. Belyakov and C.M. Gourlay, Heterogeneous Nucleationof bSn on NiSn4, PdSn4 and PtSn4, Acta Mater., 2014,71(June), p 56-68

46. W.C. Winegard, S. Majka, B.M. Thall, and B. Chalmers,Eutectic Solidification in Metals, Can. J. Chem., 1951, 29(4),p 320-327

47. W.A. Tiller, K.A. Jackson, J.W. Rutter, and B. Chalmers, TheRedistribution of Solute Atoms during Solidification ofMetals, Acta Metall., 1953, 1(4), p 428-437

48. J. Lipton, M.E. Glicksman, and W. Kurz, Dendritic Growthinto Undercooled Alloy Melts, Mater. Sci. Eng., 1984, 65(1), p57-63

49. D.H. Kirkwood, Microsegregation, Mater. Sci. Eng., 1984,65(1), p 101-109

50. J. Lipton, W. Kurz, and R. Trivedi, Rapid Dendrite Growth inUndercooled Alloys, Acta Metall., 1987, 35(4), p 957-964

51. M.A. Chopra, M.E. Glicksman, and N.B. Singh, DendriticSolidification in Binary Alloys,Metall. Trans. A, 1988, 19(12),p 3087-3096

52. D.H. StJohn, A. Prasad, M.A. Easton, and M. Qian, TheContribution of Constitutional Supercooling to Nucleation andGrain Formation, Metall. Mater. Trans. A, 2015, 46(11), p4868-4885

53. M. Easton and D. StJohn, Grain Refinement of AluminumAlloys : Part I. The Nucleant and Solute Paradigms - AReview of the Literature, Metall. Mater. Trans. A, 1999, 30(6),p 1613-1623

54. M. Easton and D. StJohn, Grain Refinement of AluminumAlloys : Part II. Confirmation of, and aMechanism for the SoluteParadigm, Metall. Mater. Trans. A, 1999, 30(6), p 1625-1633

55. M. Easton and D. StJohn, A Model of Grain RefinementIncorporating Alloy Constitution and Potency of Heteroge-neous Nucleant Particles, Acta Mater., 2001, 49(10), p 1867-1878

56. M. Easton and D. StJohn, An Analysis of the Relationshipbetween Grain Size, Solute Content, and the Potency andNumber Density of Nucleant Particles, Metall. Mater. Trans.A, 2005, 36(7), p 1911-1920

57. Thermo-Calc Software - version 3. Solder databases versions 1(TCSLD1) and 3 (TCSLD3). http://www.thermocalc.com/products-services/software/thermo-calc/. Accessed 13 Apr2016

58. K. Suganuma, S.-J. Kim, and K.-S. Kim, High-TemperatureLead-Free Solders: Properties and Possibilities, J. Met., 2009,61(1), p 64-71

59. S. Kim, K.-S. Kim, K. Suganuma, and G. Izuta, InterfacialReactions of Si Die Attachment with Zn-Sn and Au-20SnHigh Temperature Lead-Free Solders on Cu Substrates, J.Electron. Mater., 2009, 38(6), p 873-883

60. J. Harris and M. Matthews, Selecting Die Attach Technologyfor High-Power Applications, Power Electronics. http://powerelectronics.com/dc-dc-converters/selecting-die-attach-technology-high-power-applications, accessed 25 Mar 2016

61. V. Chidambaram, H.B. Yeung, and G. Shan, Reliability of Au-Ge and Au-Si Eutectic Solder Alloys for High-TemperatureElectronics, J. Electron. Mater., 2012, 41(8), p 2107-2117

62. G. Zeng, S. McDonald, and K. Nogita, Development of High-Temperature Solders: Review, Microelectron. Reliab., 2012,52(7), p 1306-1322

63. Z.X. Zhu, C.C. Li, L.L. Liao, C.K. Liu, and C.R. Kao, Au-SnBonding Material for the Assembly of Power IntegratedCircuit Module, J. Alloys Compd., 2016, 671(June), p 340-345

64. K. Sweatman, T. Nishimura, and T. Komatsu, A Nano SilverReplacement for High Lead Solders in Semiconductor Junc-tions, J. Surf. Mount Technol., 2013, 26(2), p 21-27

65. S.J. Kim, E.A. Stach, and C.A. Handwerker, Fabrication ofConductive Interconnects by Ag Migration in Cu-Ag Core-Shell Nanoparticles, Appl. Phys. Lett., 2010, 96(14), p144101-1-144101-3

66. H. Zhang and N. Lee, Reliability of BiAgX as a Drop-inSolution for High Temperature Lead-Free Die-Attach Appli-cations, J. Surf. Mount Technol., pp. 28-32, 2013

67. P. McCluskey and H. Greve, Transient Liquid Phase SinteredJoints for Wide Bandgap Power Electronics Packaging, in PanPacific Conference Proceedings, 2014, pp. 1-10

68. S.A. Moeini, H. Greve, and P. McCluskey, Reliability andFailure Analysis of Cu-Sn Transient Liquid Phase Sintered(TLPS) Joints under Power Cycling Loads, in IEEE 3rdWorkshop on Wide Bandgap Power Devices and Applications,2015, pp. 383-389

69. S.M. Choquette and I.E. Anderson, Liquid-Phase DiffusionBonding: Temperature Effects and Solute Redistribution inHigh-Temperature Lead-Free Composite Solders, Int. J. Pow-der Metall., 2015, 51(4), p 1-10

70. V. R. Manikam and K. Y. Cheong, Die Attach Materials forHigh Temperature Applications: A Review, in IEEE Transac-tions on Components, Packaging, and Manufacturing Tech-nology, vol. 1, no. 4, pp. 457-478, 2011

71. A.H.M.E. Rahman and M.N. Cavalli, Diffusion Bonding ofCommercially Pure Ni using Cu Interlayer, Mater. Charact.,2012, 69(July), p 90-96

72. H. Greve, L.-Y. Chen, I. Fox, and F. P. McCluskey, TransientLiquid Phase Sintered Attach for Power Electronics, in 2013Electronic Components and Technology Conference, 2013, pp.435-440

73. H. Greve and P. Mccluskey, Reliability of Sn based LT-TLPSJoints for High Temperature Electronic Systems, in IntegratedPower Systems (CIPS), 2014, pp. 25-27

74. V. Vuorinen, H. Yu, T. Laurila, and J.K. Kivilahti, Formationof Intermetallic Compounds between Liquid Sn and VariousCuNix Metallizations, J. Electron. Mater., 2008, 37(6), p 792-805

75. V.A. Baheti, S. Islam, P. Kumar, R. Ravi, R. Narayanan, D.Hongqun, V. Vuorinen, T. Laurila, and A. Paul, Effect of NiContent on the Diffusion-Controlled Growth of the Product


http://www.thermocalc.com/products-services/software/thermo-calc/

http://www.thermocalc.com/products-services/software/thermo-calc/

http://powerelectronics.com/dc-dc-converters/selecting-die-attach-technology-high-power-applications



Phases in the Cu(Ni)-Sn System, Philos. Mag., 2016, 96(1), p15-30

76. C.-H. Lin, S.-W. Chen, and C.-H. Wang, Phase Equilibria andSolidification Properties of Sn-Cu-Ni Alloys, J. Electron.Mater., 2002, 31(9), p 907-915

77. S. Chen, W. Lee, C. Hsu, C. Yang, H. Hsu, and H. Wu, Sn-In-Ag Phase Equilibria and Sn-In-(Ag)/Ag Interfacial Reactions,Mater. Chem. Phys., 2011, 128(3), p 357-364

78. S.-W. Cheng, C.-H. Wang, S.-K. Lin, and C.-N. Chiu, PhaseDiagrams of Pb-Free Solders and their Related MaterialsSystems, J. Mater. Sci. Mater. Electron., 2007, 18(1),p 19-37

79. G. Vassilev, N. Milchevaa, M.C. Record, and R. Mancheva,Diffusion Couple Studies of the Ni-Bi-Sn System, J. Min.Metall. Sect. B, 2012, 48(3), p 347-357

80. National Insitute of Standards and Technology, Phase Diagram &Computational Thermodynamics: Bi-Cu-Sn System, MaterialsScience and Engineering Division - National Institute ofStandards and Technology (NIST). http://www.metallurgy.nist.gov/phase/solder/bicusn.html, accessed 19 May 2016

81. W.A. Sasangka, C.L. Gan, C.V. Thompson, W.K. Choi, and J.Wei, Influence of Bonding Parameters on the Interactionbetween Cu and Noneutectic Sn-In Solder Thin Films, J.Electron. Mater., 2011, 40(11), p 2329-2336


http://www.metallurgy.nist.gov/phase/solder/bicusn.html

http://www.metallurgy.nist.gov/phase/solder/bicusn.html

Date post:	20-Mar-2019
Category:	Documents
Upload:	doanhanh
View:	217 times
Download:	0 times

Advances in Pb-free Solder Microstructure Control and ... · heterogeneities, and rapid...

Documents